Introduction: Asperger’s syndrome is generally known as a neurodevelopmental disorder. The main features of this syndrome are the lack of social interaction, non-verbal communication, unusual repetitive behavior, restricted interests, and may have an inherent talent such as mathematics, music, etc. Nonetheless, their brain structural and functional variations as compared to healthy individuals require to be well understood.
Methods: This study intends to identify differences of the task-free fMRI data in Asperger’s syndrome as compared to healthy individuals using the graph-theoretical approach. In this approach, graph local and global measures are calculated from the functional network, which estimated through taking the correlation between activities in different parts of the brain. Subsequently, the differential pattern of local and global measures in Asperger’s syndrome as compared to healthy control group is investigated. Two groups of the subjects are matched in terms of age, gender, handedness, and IQ scores.
Results: Results revealed the significant differences in local measures at temporal, amygdala, thalamus, and heschl regions. Classification of the tf-fMRI data based on the identified measures shows an accuracy of 84% to discriminate Asperger's individuals from the healthy group.
Conclusion: Accordingly, local measures extracted from the graph of the task-free functional connectivity network have a good potential for screening of Asperger's syndrome that can be used as an automatically-diagnosed method of this disorder.

Introduction
In complex systems, various activities lack coordination, making their identification challenging—a persistent issue over time. A particular cell is active in response to stimuli with specific properties so that the cell may contain information about those properties. All cells provide potent signals but cannot tell where information processing occurs in a particular brain region. The hypothesis underlying the imaging technique claims that a cell contains information about the stimulus, causing it to fire mechanically, so the cell can be interpreted as detecting that feature. Recently, this hypothesis has been challenged by researchers who have noted that most cells respond and process information at different speeds. The present study aims to answer this question: Are the results obtained from fMRI reliable or not?
The history of fMRI development has two distinct paths for research and development: the first path of physics led to the discovery of nuclear magnetism and its subsequent application in the development of magnetic resonance imaging, and the second path in physiology was to discover changes in cerebral blood flow using MRI and describe the response by blood oxygen levels. The primary goal of these experiments was not to generate new knowledge about brain activity but to replicate well-known neuroscience theories. However, in the late 19th and early 20th centuries, they used it to discover further information about the brain.

Methods
This research assesses the reliability of outcomes by consulting the perspectives of philosophers of science specializing in neuroimaging techniques. It employs philosophical argumentation and conceptual analysis, practices typical in cognitive science philosophy.
 
Results
BOLD in each brain region is tested by testing the null hypothesis of a statistically significant difference between the conditions desired by the researcher. In fMRI studies, the null hypothesis asserts that an experimental condition does not inherently influence the observed MR signal. In this way, the neural activity in the target area remains unchanged during the cognitive test. The P-value indicates the probability of observing the data under the null hypothesis. The second step compares the P-value with the predetermined significance level (α). Suppose the P-value is less than this level of significance. In that case, the data is statistically significant, and the active region of the brain is functionally substantial, allowing the researcher to reject the null hypothesis. Deborah Mayo argues that if the inferential method has very little chance of providing evidence against hypothesis H, even if H is false, then one can intuitively deny that the data are evidence for H. The stricter condition is known as the full severity principle, indicating that if Hypothesis H passes the t-test, the data provide good evidence for Hypothesis H. Mayo further states that conclusions and theories can be based on local evidence.
One way to counter Mayo's emphasis on empirical knowledge is to point out the key roles that theory plays in science, which she ignores. Theories may be criticized not by empirical tests but by showing that they are incompatible with other theories. Chalmers argues that theories are fundamental components of scientific knowledge. Therefore, they cannot be interpreted merely as heuristic guides to actual empirical knowledge developed independently of them. Likewise, Empirical knowledge cannot be justified without appealing to some theory. Similarly, Klein criticizes the testing of statistical hypotheses, such as the t-test in fMRI research. As explained, if the P-value is lower than the predetermined significant threshold, the null hypothesis is rejected, and it can be concluded that there is a statistically significant difference between the control and experimental conditions. He believes meaningful results can be obtained even when no natural effect exists. Various factors can make this happen. For example, when significant activity is observed in a brain region as a result of performing a cognitive test, this activity is likely interpreted as significant because the significance threshold was chosen freely and permissively. According to Klein, freely choosing the significance threshold harms the reliability of inferences drawn from fMRI data. The statistical nature of the inference in fMRI makes the inferences drawn from its data unreliable, as no agreement exists between the views of probability and their reliability. On the other hand, if all parts of the brain are essential for a function, the null hypothesis is always false, and the severe test of Mayo's statistical significance cannot be used.

Conclusion
The present study shows that cognitive theories do not explicitly predict brain function. How these predictions follow from their plausible theories and how psychological theories about the brain can be justified is still underdetermined at this stage. If fMRI data disproves or confirms a theory, they can be consistent with the data because these theories do not make precise predictions about brain functions. As mentioned, there are different theories about the neural circuits of fear. Each scientist makes fMRI data consistent with his mutually exclusive theory, and this, as Klein calls it, is an impasse. Additionally, it is impossible to reach a reliable conclusion by making the test conditions difficult and passing the success of the hypothesis from the severe test conditions because these features and statistical tools are full of statistical hypotheses. The problems of the theory-ladenness nature of neural images and the reliability of conclusions cannot be solved by using the science of statistics.

Compliance with ethical guidelines
There are no ethical considerations in the research relevant                   to this study.

Authors' contributions
This article is excerpted from the lead author's master's thesis in Psychology Department at Isfahan University. All three contributors collaborated in selecting the subject and refining the final manuscript.
Funding
No financial support has been received from any organization for this study.
Acknowledgments
This article is the result of a study conducted for a master's thesis at the University of Isfahan, which was not possible without the support of two respected professors. Words cannot express my gratitude towards them.
Conflict of interest
The authors declared no conflict of interest.